Short-term cold exposure induces persistent epigenomic memory in brown fat

Abstract

Deficiency of the epigenome modulator histone deacetylase 3 (HDAC3) in brown adipose tissue (BAT) impairs the ability of mice to survive in near-freezing temperatures. Here, we report that short-term exposure to mild cold temperature (STEMCT: 15°C for 24 h) averted lethal hypothermia of mice lacking HDAC3 in BAT (HDAC3 BAT KO) exposed to 4°C. STEMCT restored the induction of the thermogenic coactivator PGC-1α along with UCP1 at 22°C, which is greatly impaired in HDAC3-deficient BAT, and deletion of either UCP1 or PGC-1α prevented the protective effect of STEMCT. Remarkably, this protection lasted for up to 7 days. Transcriptional activator C/EBPβ was induced by short-term cold exposure in mouse and human BAT and, uniquely, remained high for 7 days following STEMCT. Adeno-associated virus-mediated knockdown of BAT C/EBPβ in HDAC3 BAT KO mice erased the persistent memory of STEMCT, revealing the existence of a C/EBPβ-dependent and HDAC3-independent cold-adaptive epigenomic memory.

Keywords: C/EBPβ; ERRα; HDAC3; PGC-1α; UCP1; brown adipose tissue; cold memory; mitochondria; oxidative phosphorylation; thermogenesis.

Figure 1.. STEMCT rescues the cold-intolerant phenotype of BAT HDAC3 BKO mice

(A) Schematic illustration of the experiments in (B-D and F-I). STEMCT, short-term exposure to mild cold temperature. CTT, cold tolerance test. (B-D) CTT (B and C) and mild cold exposure to 15°C (D): core body temperature (B and D) and survival (C) of control and Ucp1-cre HDAC3 KO mice. Control versus KO (n = 12, 11, in B and C; n = 12, 11, in D) The number of mice at that time point is shown in parentheses. (E) Representative image of H&E staining in BAT from control and Ucp1-cre HDAC3 KO mice housed at 22°C (n = 5, 6) or 15°C for 24 h (n = 5, 6). Scale bar, 10 μm. (F-I) CTT: core body temperature (F and H) and survival (G and I) of control and Ucp1-cre HDAC3 KO mice. Control versus KO (n = 11, 10, in F and G; n = 13, 9, in H and I). *P < 0.05, **P < 0.01, as analyzed by log-rank test (C, G and I) or Welch’s t-test (B, D, F and H). Data are represented as mean ± SEM.

Figure 2.. STEMCT induces UCP1 which is required for rescue

(A) Number of differentially expressed genes (fold-change > 1.5 & FDR < 0.01) from Ucp1-cre HDAC3 KO mice versus control mice under the conditions at 22°C or 1-day recovery (1dR). Upregulated genes are indicated above the y-axis and downregulated genes are below the y-axis. (B) Scatter plot of RNA-seq data comparing Ucp1-cre HDAC3 KO mice versus control mice under the conditions at 22°C or 1-day recovery. X and Y axis in log2 (reads per kilobase per million (RPKM)). (C) Relative mRNA levels of Hdac3 and Ucp1 in BAT from control and Ucp1-cre HDAC3 KO mice housed at 22°C (n = 11, 11) or 15°C for 24 h (n = 6, 6). (D) Immunoblot for UCP1 and HDAC3 in BAT from control and Ucp1-cre HDAC3 KO mice housed at 22°C or 15°C for 24 h. VIN, vinculin, as a loading control. (E and F) CTT: core body temperature (E) and survival (F) of Hdac3^f/f, Ucp1^+/+ (n = 8), Hdac3^f/f, Ucp1^−/− (n = 6), Adipoq-cre; Hdac3^f/f, Ucp1^+/+(n = 4) and Adipoq-cre; Hdac3^f/f, Ucp1^−/− (n = 11) mice under the 1-day recovery condition. The number of mice at that time point is shown in parentheses. n = 4 replicates in RNA-seq (A and B). *** P < 0.001, as analyzed by two-way analysis of variance (ANOVA) with Tukey’s test (C). *P < 0.05, ***P < 0.001 (Adipoq-cre; Hdac3^f/f, Ucp1^+/+ versus Adipoq-cre; Hdac3^f/f, Ucp1^−/−), ^##P < 0.01, ^###P < 0.001 (Hdac3^f/f, Ucp1^+/+ versus Hdac3^f/f, Ucp1^−/−), as analyzed by Welch’s t-test (E) or log-rank test (F). Data are represented as mean ± SEM.

Figure 3.. ERRα sites are activated by STEMCT

(A) Scatter plot comparing eRNA regulation upon cold-exposure between control and KO in log2 FC (15°C vs 22°C). (B) HOMER de novo motif search at eRNA induced in HDAC3 KO mice following STEMCT. TF, transcriptional factor. (C) Average GRO-seq eRNA profile at ERRα binding sites in 15°C and 22°C in KO. (D) Representative Integrative Genomics Viewer (IGV) tracks of the Ucp1 locus highlighting GRO-seq (22°C and 15°C for 24 h) and HDAC3 and ERRα ChIP-seq (22°C) data (y axis scales in brackets: reads per million; eRNA tracks feature adjusted y axis scale). (E) ChIP qPCR of ERRα at the enhancer regions of Ucp1 in BAT from control and Ucp1-cre HDAC3 KO mice (22°C and 15°C for 24 h). Ins, insulin, is used as a non-specific binding site. n = 5 per group. *P < 0.05, **P < 0.01, as analyzed by two-way ANOVA with Tukey’s test and ^#P < 0.05, as analyzed by Welch’s t-test. Data are represented as mean ± SEM. (F) Average ATAC-seq profiles of downregulated BAT HDAC3 BKO enhancers in GRO-seq, bound by ERRα, with rescued eRNAs following STEMCT (1-day recovery versus 22°C in control (left) and Ucp1-cre HDAC3 KO mice (right)).

Figure 4.. STEMCT induces ERRα coactivator PGC-1α which is require for rescue

(A) TRRUST enrichment analysis using Metascape of the upregulated transcriptional factors in Ucp1-cre HDAC3 KO mice following STEMCT. (B) HOMER de novo motif search at PGC-1α binding sites in BAT from PGC-1α-HisHA mice following a 4 h exposure to 4°C. TF, transcriptional factor. Data of PGC-1α ChIP-seq are used dataset, GSE213601. (C) Venn diagram showing overlap of peaks between ERRα and PGC-1α. (D) Representative IGV tracks of the Ucp1 locus highlighting PGC-1α and ERRα ChIP-seq data (y axis scales in brackets: reads per million). (E) Representative IGV tracks of the Ppargc1a (Pgc-1α) locus highlighting GRO-seq (22°C and 15°C for 24 h) data (y axis scales in brackets: reads per million). (F) Relative mRNA levels of Pgc-1a in BAT from control and Ucp1-cre HDAC3 KO mice housed at 22°C (n = 11, 11) or 15°C for 24 h (n = 6, 6). (G and H) CTT: core body temperature (G) and survival (H) of control and Ucp1-cre HDAC3/PGC-1α dKO mice under the condition at 22°C (n = 11, 11) and 1-day recovery (n = 13, 13) condition. The number of mice at that time point is shown in parentheses. **P < 0.01, ***P < 0.001, as analyzed by two-way ANOVA with Tukey’s test (F). **P < 0.01 (22°C: Hdac3^f/f, Pgc-1_α^f/f versus Ucp1-cre; Hdac3^f/f, Pgc-1_α^f/f), ^#P < 0.05, ^###P < 0.001 (1-day recovery: Hdac3^f/f, Pgc-1_α^f/f versus Ucp1-cre; Hdac3^f/f, Pgc-1_α^f/f) by Welch’s t-test (G). NS, not significant. Data are represented as mean ± SEM.

Figure 5.. Memory of STEMCT protects for 7 days

(A) Schematic illustration of the experiments in (B-G). (B-G) CTT: core body temperature (B, D and F) and survival (C, E and G) of control and Ucp1-cre HDAC3 KO mice under the conditions at 7- (n = 12, 11), 14- (n = 11, 11) and 30-day (n = 9, 11) recovery. The number of mice at that time point is shown in parentheses. (H) Principal component analysis (PCA) plot from bulk RNA-seq experiments in control and Ucp1-cre HDAC3 KO mice under the conditions at 22°C, 1 and 7-day recovery. (I) Immunoblot for UCP1 in BAT from control and Ucp1-cre HDAC3 KO mice under the conditions at 22°C, 15°C for 24 h and 1– 7- and 30-day recovery. VIN as a loading control. Right panel is shown quantification for UCP1 from control and Ucp1-cre HDAC3 KO BAT (n = 2 per each condition). (J) ATAC-Seq profiles of downregulated HDAC3 KO enhancers in GRO-seq, bound by ERRα, with rescued eRNAs following STEMCT (7-day recovery versus 22°C in control (left) and Ucp1-cre HDAC3 KO mice (right)). (K) Relative mRNA levels of Ucp1 in BAT from control and Ucp1-cre HDAC3 KO mice under the conditions at 22°C (n = 11, 11) and 30-day recovery (n = 8, 8). *P < 0.05, **P < 0.01, as analyzed by Welch’s t-test (B, D and F), log-rank test (C, E and G), or two-way ANOVA with Tukey’s test (K). Data are represented as mean ± SEM.

Figure 6.. C/EBPβ, but not PGC-1α, remains induced 7 after STEMCT and activates UCP1 enhancer activity

(A) Relative mRNA levels of Cebpb and Cebpa in BAT from control and Ucp1-cre HDAC3 KO mice housed at 30°C for 7 days (n = 8, 8), 22°C (n = 11, 11) or 15°C for 24 h (n = 6, 6). (B) Schematic representation of biopsy in human supraclavicular BAT, created using BioRender. (C) Relative mRNA levels of UCP1, CEBPB and CEBPA in human BAT at thermoneutrality (TN, 26–28°C) and cold exposure (CE, 20C). n = 16 per group. (D) Representative IGV tracks of the Ucp1 locus highlighting GRO-seq (22°C and 15°C for 24 h) and C/EBPβ ChIP-seq data (y axis scales in brackets: reads per million; eRNA tracks feature adjusted y axis scale). (E) eRNAs at C/EBPβ-bound enhancers in BAT HDAC3 KO (15°C versus 22°C). (F) ChIP qPCR of C/EBPβ at the enhancer/promotor regions of Ucp1, Ppargc1a and Cebpb in control and Ucp1-cre HDAC3 KO mice housed at 22°C or 15°C for 24 h. n = 5 per group. (G and H) Immunoblot for C/EBPβ in nuclear extracts of BAT from control and Ucp1-cre HDAC3 KO mice under the conditions at 22°C, 15°C for 24 h and 1-, 7- and 30-day recovery. H3 as a loading control. LAP, liver-enriched activating protein. LIP, liver-enriched inhibitory protein. The band intensity of C/EBPβ protein in Figure S6J was quantified using ImageJ software (H). (n = 3 per group). (I) ATAC-Seq profiles bound by cold-induced C/EBPβ following STEMCT (1- or 7-day recovery versus 22°C in control and Ucp1-cre HDAC3 KO mice). Cold-induced C/EBPβ binding is used ChIP-seq peaks induced at 4°C compared to 22°C. *P < 0.05, **P < 0.01, ***P < 0.001, as analyzed by two-way ANOVA with Tukey’s test or Welch’s t-test. *P < 0.05 (versus control at 22°C), ^#P < 0.05 (versus Ucp1-cre HDAC3 KO mice at 22°C) by Welch’s t-test (H). Data are represented as mean ± SEM.

Figure 7.. C/EBPβ is indispensable for the memory of STEMCT

(A) Relative mRNA levels of Cebpb in BAT from control and Ucp1-cre HDAC3 KO mice housed at 15°C for 24 h (n = 5, 6) after direct delivery of AAV-shControl or AAV-shCebpb for 2 weeks. (B) Immunoblot for C/EBPβ in BAT from control and Ucp1-cre HDAC3 KO mice housed at 15°C for 24 h after direct delivery of AAV-shControl or AAV-shCebpb for 2 weeks. VIN, vinculin, as a loading control. LAP, liver-enriched activating protein. (C) Schematic illustration of the experiments in (D and E). (D and E) CTT: core body temperature (D) and survival (E) of control and Ucp1-cre HDAC3 KO (KO) mice with shControl or shCebpb under the condition at 22°C (shControl; n = 10, 10, shCebpb; n = 11, 8) and 7-day recovery (n = 10, 10) condition. The number of mice at that time point is shown in parentheses. *P < 0.05, ***P < 0.001, as analyzed by Welch’s t-test (A and D). D, Control, shControl at 22°C versus KO, shControl at 22°C. ^#P < 0.05 (Control, shCebpb at 22°C versus KO, shCebpb at 22°C or KO, shCebpb upon 7-day recovery) by Welch’s t-test. NS, not significant. E, NS by log-rank test (KO, shCebpb at 22°C versus KO, shCebpb upon 7-day recovery). Data are represented as mean ± SEM.